ansys advantage special issue: oil and gas/media/ansys/corporate/... · · 2017-04-05in a drill...

Excellence in Engineering Simulation 2014

ADVANTAGETM

SPECIALISSUE:Oil and Gas

Product reliability and Performance Accelerate product reliability and performance with systematic use of engineering simulation. By Ahmad H. Haidari, Global Industry Director, Energy and Process Industries, ANSYS, Inc.

The oil and gas industry’s sharp focus on safety and reliability is based on the economics of oil and gas production, maintenance and unscheduled downtime — and along with that comes environmental stewardship and protecting human life. As project costs and complexity

increase, companies develop continuous-improvement processes to reduce equipment and product failure to increase operational reliability at drilling, production and processing sites. This industry strives for zero accidents and, therefore, undertakes extraordinary measures to avoid loss to human life, environment and capital. Equipment reliability is the key to drive operational excellence and profitability, reduce resource waste, eliminate unnecessary downtime, decrease over-design and unplanned maintenance as well as non-productive time, and create smart intervention and preven-tion strategies. In fact, the industry spends a large part of its R&D budget on developing reliable products. The challenge is to develop these products for real-life conditions, environments that are often impossible to replicate with experiment.

Many variables must be considered when designing a new prod-uct. Because it isn’t possible to test and prototype every permuta-tion, some designers over-design parts or focus on the 20 percent of equipment and processes that they consider the main source of a problem. For existing equipment and facilities in service, some rely heavily on inspection and data gathering, incorporat-ing historical data. These practices miss the mark: They can lead to poor product design and may cost millions of dollars for field testing along with related maintenance and inspection expenses. Furthermore, these procedures provide little insight that can be applied systematically and/or scaled to developing novel

concepts and ensuring product performance during off-design conditions. To guarantee product performance and compliance with industry standards and regulations, engineers must apply reliability practices to a wide range of equipment, products and processes — for example:• Pumps • Drill systems• Actuators • Pressure and flow control devices• Sensors • Controlling software• Valves • Electronic and electrical devices• Transmitters • Wireless signals

overview

2 in-depth SolutionTechnip automates evaluation of 20,000 simulation runs to ensure that subsea pipe structures can survive worst-case scenarios.

TABLE OF CONTENTS

6Pipe dream becomes realityAccurate simulation improves reliability and ensures cost-effective deployment of pipe strings in oil wells.

10a Perfect fitResearchers develop an automated process for optimizing marine structural components.

14designing for real-world repairsLinear and nonlinear structural analyses improve pipeline repair using composites materials.

ANSYS ADVANTAGE I 2014

http://ansys.com/Business+Initiatives/Product+Integrity+Through+Robust+Design+Optimization

ANSYS.COM SPECIAL ISSUE: OIL AND GAS 1

Independent of how data is gathered (probability, historical data, etc.) or what analysis method is employed, the business impact of product failure is often much more than the initial cost of get-ting the design right, not to mention the price of understanding root causes to accurately predict failure. Common failure modes are equipment-specific and often depend on process conditions — friction, corrosion, erosion, fatigue, thermal stress, vibration, etc. In addition, equipment performance is influenced by the choice of material, operating environment, manufacturing pro-cesses (including variation in process and operating conditions), and underlying structural, fluid mechanics, electrical and chem-ical considerations. Detailed engineering simulation that considers all applicable physics complements current reliability and product design practices. It enables parametric, systematic analysis of compo-nents and key subsystems in real-life situations. Components and subsystems can be designed, analyzed and validated — across a broad range of physics and operating conditions — to account for uncertainties via computational techniques for vir-tual prototyping and experimentation. One key tool for success is an integrated framework that enables variation in design across multiple dimensions (geometric scale, physics and domain) and facilitates cross-functional engineering collaboration. The ANSYS framework and methodologies can help organiza-tions to evolve the use of engineering simulation from a single, siloed design validation practice to a procedure that drives opti-mization in a world of design input uncertainty. ANSYS solutions combine robust design methodology centered on parametric analysis, design exploration, goal-driven optimization and prob-abilistic optimization with comprehensive solutions, all applica-ble to design of advanced material systems, fluid–mechanical systems, electric machine and drive systems, and fluid–thermal systems. ANSYS solutions are ideal for a range of oil and gas industry initiatives. By leveraging the software, teams can

streamline product design for reliability, develop reliable subsea systems, and integrate solutions for marine, subsea and offshore structures as well as midstream, downstream, LNG, and FLNG equipment design and optimization. These solutions offer a high degree of certainty that the design is right the first time. This level of high-fidelity analysis and design also benefits flow assur-ance, wellbore and near-wellbore equipment reliability and projects. Robust, accurate and proven engineering simulation solutions can be beneficial for evaluation of new concepts and the design of new equipment and facilities. For equipment already in opera-tion, simulation is a powerful tool for identifying failure mode and root causes, driving maintenance schedules, increasing throughput, troubleshooting and establishing fitness for service. This special issue of ANSYS Advantage includes an array of cus-tomer case studies and industrial application articles that might inspire you to take even more advantage of engineering simula-tion. In “In-Depth Solutions,” for example, Technip describes how it automates ANSYS software to evaluate over 20,000 simu-lation runs that ensure subsea jumper pipe structures can survive worst-case scenarios. In “Pipe Dream Becomes Reality,” Schlumberger engineers use nonlinear FEA simulation of BHA in a drill string to eliminate lateral buckling. Similar optimiza-tion and simulation solutions are used for marine structures in the study titled “A Perfect Fit.” Other articles demonstrate the benefits of ANSYS comprehen-sive capabilities. The accelerating use of ANSYS engineering simulation solutions throughout the energy industry helps organizations to develop reliable products, increase product performance and reduce environmental impact. They rely on a range of well-established comprehensive solutions covering structural, fluid, electromagnetic and hydrodynamic behavior coupled in an integrated framework. Most important is that by using ANSYS software, oil and gas companies can reduce risk while meeting corporate project and production objectives.

18designing Solid compositesEmploying ANSYS Workbench workflow streamlines simula-tion of solid composites.

21Pushing the envelopeCFD simulation contributes to increasing the operating envelope of a centrifugal compressor stage.

.26 raising the StandardsFluid–mechanical simulation can help prevent offshore disasters by supporting development of more effective structural standards.

By Esen Erdemir-Ungor, Design Specialist, Technip, Houston, U.S.A.

J umpers are piping components of subsea oil production systems that connect one structure to another,

such as for linking satellite wells to a manifold, the platform or other equip-ment. Designing these very important components is diffi cult because both of the connection points are free to move —within allowable limits — due to thermal expansion, water currents and other fac-tors. Jumper designers need to evaluate every possible combination of movement, expansion and rotation to determine which combination applies the most stress to the jumper, then design the jumper to withstand it.

Technip recently designed four jump-ers, each connecting a pipeline end termi-nation (PLET) — the end connecting point of a pipeline — to the manifold of a pro-ducing well or another PLET. Technip is a world leader in project management, engi-neering and construction for the energy

industry. With facilities in 48 countries, the company operates a fl eet of special-ized vessels for pipeline installation and subsea construction.

loaDs on thE JumpErUndersea pipelines are governed by strict codes developed to ensure pipeline integ-rity to prevent an oil spill. The jumper needs to withstand loads applied to both ends of the pipe while keeping stress in the jumper within the limits specified by the code.

When oil or gas is transported in the pipeline, the pipeline undergoes thermal expansion, and this expansion is trans-mitted to the jumper. In this Technip application, thermal expansion was cal-culated to be a maximum of 40 inches in the x-axis and 30 inches in the z-axis. Further displacements of up to 2 inches in the x-, y- and z-axes were possible due to variation when the position of

the structures was measured and when the jumper was cut and assembled to its fi nal size. Rotations of up to 5 degrees in either direction in the x- and z-axes were also possible. The net result was a total of three displacements and two rotations on each end of the jumper that needed to be considered at each extreme of its range of motion. To fully understand every load case that could be applied to the jumper, it’s necessary to consider every possible combination of these 10 different vari-ables, a total of 1,024 load cases.

Technip engineers had to take into account variability in the position of the PLET and manifold. There is a target loca-tion for the two structures, but the posi-tion can vary within the project-specifi ed target box. As a result, the length of the jumper can be anywhere from 900 inches to 1,500 inches; furthermore, the gross angle of the jumper with respect to the PLET and manifold also can vary. This

offSHore

In-DepthSolution

Technip automates evaluation of 20,000 simulation runs to ensure that subsea pipe structures can survive worst-case scenarios.

2 ANSYS ADVANTAGE I 2014

ANSYS.COM

a relatively small number of load cases that they believe will generate the high-est level of stress. But operators of wells and pipelines are becoming much more sensitive to potential hazards. In this project, the customer asked that every single load case be evaluated to make cer-tain that the jumper could withstand the absolute worst case. Just a few years ago, such a task would take so long that organ-izations would rule it out for production jobs. But recent advances in optimiza-tion tools now make it possible to rapidly evaluate large numbers of design cases to ensure robustness.

Exploring thE DEsign spacEIn this project, the first step was to cre-ate a simple jumper model in ANSYS DesignModeler based on a previous design. Engineers created three design

parameters to define the geometry of the jumper that could be varied to improve its performance. Parameters included the length of two vertical and one horizontal sections of pipe that constitute the core of the jumper (geometric parameters) as well as three displacement and two rota-tion parameters at each end of the jumper (mechanical parameters), with two possi-ble values representing each extreme end of its range of motion.

As the first step of the design proc-ess, engineers set up a short simulation run to explore the design space. They selected a previous design as the start-ing point, and the geometric design parameters were allowed to vary over a limited range in increments of 1 foot. Engineers used the Design Points option in ANSYS DesignXplorer to select a subset of about 200 load cases. They

� Parameters were allowed to vary during optimization. The diagram shows loads that potentially can be applied to the jumper. Ten variables were applied to the remote displacements.

3 displacements + 2 rotations

3 displacements + 2 rotations

Pho

to c

ou

rtes

y te

chn

iP.

Using conventional analysis tools, it would be impossible for an engineer to solve this many load cases within a normal design cycle.

gross angle is important because it deter-mines the angle at which thermal expan-sion is applied to the jumper. The position of the PLET and manifold are measured prior to jumper installation. The jumper is then cut and welded to the length and angle determined by the measurements just before it is installed. The engineer-ing team addressed these variations by considering four different scenarios for the jumper: maximum length, minimum length, maximum gross angle and mini-mum gross angle. So a complete evalua-tion requires that the 1,024 load cases be evaluated for each of these four scenarios, resulting in a total of 4,096 load cases for each jumper design.

Using conventional analysis tools, it would be impossible for an engineer to solve this many load cases within a nor-mal design cycle. The standard practice has been for experienced engineers to use their judgment and instinct to pick out

SPECIAL ISSUE: OIL AND GAS 3

http://www.ansys.com/Products/Workflow+Technology/ANSYS+Workbench+Platform/ANSYS+DesignModeler

offSHore

� Total deformation (top) and maximum combined stress (bottom) at true scale

� The design of experiments capability in ANSYS DesignXplorer helped the simulation software to run thousands of load-case steps. Actual mechanical parameter ranges are shown.

created a table with these parameters within the DesignXplorer optimiza-tion tool. A Technip engineer gave the Update command to solve the model for every combination of values in the table. The first design point, with the first set of parameter values, was sent to the parameter manager in the ANSYS Workbench integration platform. This drove the changes to the model from CAD system to post-processing.

DesignXplorer used parametric per-sistence to reapply the setup to each com-bination of parameters while file transfer, boundary conditions, etc., remained per-sistent during the update. The new design point was simulated, and output results were passed to the design-point table where they were stored. The process con-tinued until all design points were solved to define the design space. The outputs of each simulation run included the mini-mum and maximum bending stress, shear stress, axial stress and combined stress within the jumper. Technip engineers examined the results, looking particu-larly at the sensitivity of the outputs with respect to design parameters and whether their variation with respect to the design parameters was linear or nonlinear.

DEtErmining thE Worst-casE scEnarioAs the second step, engineers fixed the mechanical parameters at the values that provided the worst results in the previ-ous step with the goal of obtaining the geometric parameter set that could with-stand the worst load combinations. Once the mechanical parameters were set at the current worst case (obtained from the first step), then the geometric parameters were allowed to vary over a greater range. Technip created a design-point table using the default settings in the design of experiments. Engineers employed goal-driven optimization for which the primary goals were that the stresses mentioned previously would not exceed allowable values. At the end of the second step, a set of geometric parameters that do not fail under the current worst-case scenario was obtained.

The third step confirmed that the optimized geometric parameter set would not have stresses higher than the allowable values under any pos-sible load combination. Technip engi-neers created a design-point table using


ANSYS.COM

Scaling Design ParametersBy Mai Doan, Senior Application Engineer, ANSYS, Inc.

Technip’s customer wanted a rigorous study of every possible combination of parameters, and ANSYS DesignXplorer was up to that task. However, many companies employ DesignXplorer to study the design space with as few solved design points as possible. Advanced DOE and optimization algorithms within this tool enable users to choose combinations of parame-ters that extract the maximum amount of information with minimum resources. Response surfaces (also known as metamodels) interpolate between the solved design points. If, for example, peak loads or optimal designs are predicted between solved design points, these can easily be verifi ed on an as-needed basis. Using automated refi nement and adap-tive optimization, DesignXplorer focuses solver resources in the areas of the design space that are most likely to yield valuable results. � Response surfaces show the relationships between design parameters

and design performance.

� Jumper installation This capability will provide Technip with the signifi cant competitive advantage of being able to prove to clients that its designs can withstand worst-possible conditions.

the two possible extreme values (mini-mum and maximum) for each mechani-cal parameter while fi xing the geometric parameters at the values obtained in the previous step. Since there are 10 mechanical parameters, this resulted in 1,024 (210) load cases. The Custom Design Point table option was used to import the 1,024 determined load

cases. Engineers monitored the design-of-experiments runs, and if, for any load combination, the allow-able values were exceeded, the design-point update was stopped, then the mechanical parameters that produced high stress were set to the new worst-case scenario. This started the iterativeprocess between the second and the third steps. When all the runs in the third step were completed success-fully, so that the allowable values were not exceeded within the pipe and the reactions at the ends of the jumper did not exceed connector limits, engineers moved onto the fourth step.

The 1,024 load combinations for each of the other three scenarios discussed ear-lier were run using design of experiments for step four. When all the design criteria were met for all 4,096 possible load com-binations, engineers deemed the opti-mized parametric set successful, and the design for the fi rst jumper was fi nished.

For the second, third and fourth designs, Technip engineers started with the optimal design that had been deter-mined for the fi rst jumper. They ran this design against the 4,096 load cases for each of the other jumpers. The maximum stresses were not exceeded on the last three jumpers — for each jumper design, engineers ran only the 4,096 cases needed

to prove that the design could withstand every possible load case.

Variations in operating conditions may create uncertainty in subsea pipe structural design. Using parametric exploration and optimization tools from ANSYS, engineers checked the struc-tural performance and integrity of these four jumpers over about 20,000 simula-tion runs. This capability will provide Technip with the signifi cant competitive advantage of being able to prove to cli-ents that its designs can withstand the worst-possible conditions encountered under the sea.


A s the search for oil and gas progresses into increasingly deeper waters with exposure to higher downhole pressures and temperatures, accurate prediction of the complex stress state in oil well pipe strings is critical to ensure that operations are carried out safely and efficiently.

Accurate nonlinear structural mechanics simulation makes it possible to predict the behavior of the pipe string as it undergoes complex buckling that often results from fluid injection opera-tions. The latest simulation methods enable engineers to design pipe strings and downhole tools with the capabilities to handle the more challenging wells being drilled today.

Schlumberger, the world’s leading supplier of technology solutions to the oil and gas industry worldwide, performs well tests to obtain important measurements, such as flow rate and

downHole

pipE DrEam BEcomEs rEalitYAccurate simulation improves reliability and ensures cost-effective deployment of pipe strings in oil wells.

By Jim Filas, Technical Advisor, Schlumberger Rosharon Testing and Subsea Center, Rosharon, Texas

bottomhole pressure, to characterize petroleum reservoirs. During well testing operations, fluids may be pumped under high pres-sure into the wellbore to stimulate the formation (rock around the borehole).

To conduct a well test, a bottomhole assembly (BHA) consist-ing of specialized tools and measuring instruments is conveyed downhole on the end of a pipe string and lowered into the well casing. A packer on the BHA’s lower end expands within the well to isolate the interior of the pipe string from the annulus between the pipe string and casing. The packer has a smooth internal seal bore that accommodates a seal on the bottom of the BHA. This arrangement permits vertical movement of the lower end of the pipe string while maintaining a seal with the internal diameter of the packer to allow for thermal expansion and contraction of the pipe string during the well test. The lowest stiffness tubular

Pho

to c

ou

rtes

y sc

hLu

MBe

rGer

.


ANSYS.COM

The latest simulation methods enable engineers to design pipe strings and downhole tools with the capabilities to handle the more challenging wells being drilled today.

� Representative well test string

� Hydraulically induced bottom force

� Bending moment in helically buckled pipe predicted by analytical solution

Annulus pressure, p0

Tubing pressure, pi

Stinger

Seal bore area

Aseal

Atube

Ap

Fh (pi >p0 )

Helix

Contact points

Dimensionless distance above packer

Dim

ensi

onle

ss m

omen

t

member in the BHA is usually the stinger, whose lower end is fi tted with the afore-mentioned seal assembly and inserted into the packer seal bore. The stinger typ-ically extends out of the top of the packer a distance of 30 feet or more and is joined to the drill-stem test (DST) tools above, which in turn are joined to the major pipe string to the surface.

The lower end of the stinger is free to move vertically, so when the inter-nal string pressure exceeds the pressure in the annulus, a hydraulically induced upward force is applied to the bottom of the stinger. Such conditions exist when fl uid is pumped downhole and forced into the formation, such as during hydrau-lic fracturing or acidizing operations. The resulting pressure wave caused by fi ring perforating guns [1] can also apply similar upward hydraulic forces.

These upward forces on the bottom of the stinger can cause the BHA and pipe string to helically buckle inside the cas-ing. When helical buckling occurs, as long as the elastic limit of the tubing is not exceeded, the string components will return to their initially straight condition when the pressure diff erence is removed. However, if the elastic limit is exceeded, permanent corkscrewing of the BHA will result. The stinger, in particular, may become jammed within the casing, pre-venting the retrieval of the string from the hole and resulting in the loss of expensive tubular assets. In the worst case, this can cause a safety hazard.

prEvious DEsign mEthoDsAnalytical methods have traditionally been used to predict the buckling behav-ior of the pipe string. First developed by Arthur Lubinski in the 1960s and refi ned by others in the ensuing decades, these methods not only take into account the hydraulic force acting on the bot-tom of the string but also the infl uence of internal and external tubing pres-sure on the lateral stability of the pipe string. Analytical methods are limited to very simple geometries and typically do not account for geometric nonlinear-ities, such as large defl ections and com-plex contact between the pipe and casing, so they are unable to accurately predict complex combinations of eff ects found in the real world. In addition, existing ana-lytical solutions are limited in their abil-ity to accurately predict the post-buckled shape of the packer’s stinger in the region where it exits the packer seal bore and before the string fully forms into a helix.

Prior use of fi nite element analysis (FEA) to study helical buckling required engineers to build large models compris-ing solid elements. A perfectly symmetri-cal and straight pipe will not buckle in a numerical simulation, so it is necessary to apply small, random lateral loads to simulate imperfection, and then incre-mentally increase the bottom load to induce buckling. Lateral defl ection is con-strained after the pipe contacts the casing wall, and equilibrium is re-established as the pipe forms into a stable helical shape. Modeling a substantial length of the well test string using solid elements and solv-ing a suffi cient number of iterative steps to explore the full load range of interest requires enormous computing resources and wall clock time.


Using BEAM188 Elements for Nonlinear SimulationThe BEAM188 element has six degrees of freedom at each node and is based on a fi rst-order shear deformation theory that is well suited for linear and nonlinear problems with large rotations and strains. Cross sections remain plane and undistorted after deformation, but they are not required to remain perpendicular to curvature.

Once the lower end of the string buckles and the bottom load is further increased, the model remains continuously unstable as new helix coils are formed progressively higher in the string. For such a highly nonlinear problem, aggressive stabilization con-trol is needed to maintain convergence as the solution progresses. Nonlinear stabilization using pseudo-viscous damping was used to provide the necessary control. Dampers, with appropriate damping coeffi cients, are attached to each node in the system. At the onset of instability, the integration increment is reduced when divergence of the solution is detected. The damping forces act in the opposite direction of the nodal displacements to enable the solver to obtain a converged solution during what would otherwise be an unstable phase of the simulation.

Therefore, the FEA model must be shortened considerably to avoid an impractically large number of degrees of freedom. Limiting the model length in this manner requires imposing boundary conditions, either displacements or forces, where the model is terminated. These boundary conditions are not known before the problem is solved. Applying assumed, imprecise boundary conditions can substantially reduce the accuracy of the simulation.

An alternative approach is to use 3-D beam elements to model a long section of the pipe string in combination with a concentric contact surface to represent the casing constraint. A line element model can be made long enough to represent a signifi cant portion of the pipe string while keeping the model size manageable. However, traditional beam elements based on classical Euler–Bernoulli theory do not work well in this type of model because of inherent restrictions. One of these restrictions is the requirement for the beam cross sections to remain perpen-dicular to curvature, which prevents the helix from developing after the initial buckling load has been reached.

It is important to more accurately predict buckling and the resulting stress state in the well test string.

� Schematic of fi nite element method

� Bending moment in helically buckled pipe predicted by FEA *For 3.5-inch continuous string in 9.625-inch casing

*

downHole


ANSYS.COM

nEW simulation approachSchlumberger has developed a new approach employing ANSYS Mechanical BEAM188 elements. The team models a sufficiently long portion of the well test string so that the upper boundary condition remains above a sufficiently long portion of the fully developed helix such that the upper boundary condition does not affect the region of interest near the packer. The model is solved as a large deflection, nonlinear buckling problem along a load path that progresses from initial instability and casing con-tact, to formation of a fully formed helix, and finally to a bot-tom load magnitude that corresponds to field conditions of prac-tical interest.

To compare the FEA approach with the analytical solution, Schlumberger engineers ran FEA cases for uniform strings of vary-ing diameters constrained by different-sized casings. Unlike ana-lytical solutions, the FEA simulation fully accounts for geomet-ric nonlinearity. Next, the models were enhanced to represent a realistic BHA having regions with different diameters and flexural stiffness. Such an analysis is not practical using a simplified ana-lytical solution.

The new FEA approach provides a more accurate method of simulating helical buckling in a well test string sealed in a packer. For the case of a uniform stinger, there is excellent agreement between established analytical solutions and FEA within the fully developed helix. In the region between the packer and fully developed helix, FEA predicts higher bending stresses than exist-ing analytical solutions. Furthermore, the bending stress in the stinger just above the packer was found to vary somewhat depend-ing upon problem geometry, indicating the influence of geometric nonlinearity on the solution. The discrepancy in the region before the helix is fully developed is attributed to simplifying assump-tions inherent in the derivation of the analytical solutions. A non-linear finite element solution is not restricted to these simplifi-cations. The higher accuracy of the new FEA method will make it possible to design BHAs that ensure safe operation in deeper wells, which can help satisfy the world’s need for oil and gas.

Footnote

[1] Perforating gun: downhole device containing explosive charges used to perforate sections of the casing below the packer to allow oil and gas to enter the lower portion of the wellbore

� Shape of helically bucked well test string from simulation

The approach provides a more accurate method of simulating helical buckling in a well test string sealed in a packer.

The higher accuracy of this FEA method makes it possible to design BHAs that ensure safe operation in deeper wells, which can help satisfy the world’s need for oil and gas.


http://www.ansys.com/Products/Simulation+Technology/Structural+Analysis

researchers develop an automated process for optimizing marine structural components.

By Jouni Lehtinen, Research & Development Engineer, MacGregor Dry Cargo, Kaarina, Finland, andSami Pajunen, Associate Professor, and Ossi Heinonen, Researcher, Tampere University of Technology, Tampere, Finland

marine

apErfEctfit


ANSYS.COM

n early all marine structural com-ponents are custom designed for a specific application. It is also a

fact that the highly competitive shipping market has no room for slack. Therefore, advanced ship builders and cargo sys-tem suppliers must optimize their struc-tural designs to meet specific application needs. The bottom line: The structure’s materials must be in the exact right place — where they best support the needs of the cargo system and enable efficient marine cargo transports. An optimized steel structure with no excess weight translates into optimized and flexible space for transported cargoes.

To address such issues accurately and efficiently, MacGregor Dry Cargo’s engi-neering department and researchers at Tampere University of Technology devel-oped an automated solution for optimiz-ing marine structures; at the same time, the solution ensures that the structures are able to handle the required operating loads. This is done by means of a script that drives an ANSYS template file to perform a finite element analysis (FEA) on a series of design points. The results are used to construct a response surface model (RSM) of the design space. The RSM is reviewed to identify the most efficient design. This process also improves the reliability of the design by reducing the potential risk of design errors.

DEvEloping nEW, EfficiEnt DEsign mEthoDsMacGregor offers integrated cargo flow solutions for maritime transportation and offshore industries. The competence center for MacGregor’s Dry Cargo busi-ness line has a long history of cooperation with Tampere University of Technology, Finland’s second-largest university in engineering sciences, for research and development of new design proc-esses and tools.

In this particular marine compo-nent application, the team optimized the design to meet specific customer require-ments. The cargo profile dictates the basic parameters of the ship’s hull design. Within these constraints, the hull should be as light as possible in order to minimize material costs and also to keep the weight of the hull as low as possible. Any weight that is saved in the hull and cargo sys-tem design can be used for the benefit of the payload.

Using a standard design in this appli-cation — one that isn’t optimized to the application — would have increased the amount of material used for the prod-uct with no additional value for the cus-tomer. Reusing previous designs with similar specifications also can be diffi-cult, because many existing products were customized for project-specific requirements. Furthermore, the tradi-tional approaches do not take advantage of technological advances in paramet-ric design. Another solution for cus-tomer-specific optimization — such as simple design rules based on mathe-matical functions — does not take into account the detailed geometry of the structure, so designs created using this method are less than optimal. The most common method, conven-tional FEA, has the ability to accurately predict the performance of any sin-gle design. However, manual design optimization with FEA requires that a

skilled analyst individually study many different models. The high cost and long leadtimes of this process drive up engi-neering costs. Manual design optimi-zation takes too long to use during the tendering stage, when customers come to MacGregor for a price quote and an initial design to be produced in short turnaround. Finally, assigning experi-enced analysts to repetitive work is often not the best use of resources.

An automated process to optimize marine structures for any application has to address the dimensions and load-ing of the structure, factors that may vary drastically from project to proj-ect. The main components of cargo han-dling equipment, in this case, are the top plate, longitudinal and transverse support beams, top plate stiffeners and bottom plate. Designers begin the proc-ess by creating a parametric mid-sur-face geometry model in SolidWorks® CAD software. The team uses symmetry

With the automated optimization solution, MacGregor has a tool that optimizes the process more accurately and efficiently than before.

� A ship outfitted with a cargo handling system from MacGregor


� Convergence of optimization sequences based on two different initial designs

This process improves design robustness by reducing the potential for design error.

� Half of the structure with part of top plate removed to reveal structural members

� The most critical areas of the top plate, marked in red, are checked for buckling during the optimization process.

Bottom plate

Plate stiffeners

Longitudinal support beam

Transversal support beam

Top plate

Iteration Cycle

marine

to reduce the model to half of the struc-ture. To employ this model, the cus-tomer provides the main dimensions of the structure during the tendering pro-cedure, and the team enters these values as parameters into the surface model. Designers then parameterize the mate-rial thicknesses of the model as the key design variables to be optimized during the automated process.

automating fEa moDEl crEationAn ANSYS Workbench template file that contains ANSYS Parametric Design Language (APDL) commands automat-ically meshes the model using pre-defined meshing control settings. The team loads the structure with multi-ple uniform pressures as determined by the structural codes, and the structure is supported at designated points on the edges. The loads mainly cause compres-sive stress in the top plate, tensile stress in the bottom plate, and shear stresses on the support beams. The template file generates the load, support and mate-rial property definitions. The supports are defined with an APDL command that determines the displacement and rota-tion of specified nodes. Loads are defined using another APDL command to apply a surface force. The team uses the Named Selections feature to select the nodes and elements for applying the supports and loads. For example, the edges in the symmetry plane are defined as Named Selections and used for locating the sup-ports. Named Selections are also used to define surfaces with the same material thicknesses in selection groups, so the


http://www.ansys.com/Products/Workflow+Technology/ANSYS+Workbench+Platform



ANSYS.COM

� Optimization process

MacGregor can respond to a customer inquiry with a speed and accuracy that has not existed before, and with a design that has been optimized for each specific application.

…

thickness of each element in selection groups can be defined with the APDL commands SECTYPE and SECDATA.

Researchers selected MATLAB® scripts to control the optimi-zation. They used data from previous projects to select an ini-tial design point that reduces the number of iterations required to reach the optimal solution. The scripts generate a D-optimal or Latin hypercube sampling (LHS) design of experiments (DOE) model centered on the initial design point and then drive ANSYS Workbench to create all of the models required to evaluate each run in the DOE model. The results are used to produce an RSM that approximates the complete design space based on the results of a relatively small number of FEA iterations.

optimizing thE structurEThe scripts then scan the responses of each output variable in the RSM in a region of interest around the initial point using the simplex algorithm and determine the optimal design. Critical areas in each structural member group are defined as Named Selections so their stresses can be accessed with APDL post-processing commands. The team can easily access the mass of the structure through APDL commands. The optimization min-imizes the mass of the structure with respect to constraints based on regulatory codes. The top plate is mostly under compressive and shear stress and, therefore, is constrained against buckling. The beam webs are almost entirely under shear stress and con-strained against bucking. The stresses in the bottom plate are constrained so that the material will not yield. The design vari-ables in the optimal design are used to generate another FEA iter-ation to confirm the RSM prediction.

The group tested the automated optimization method by applying it to a stiffened plate structure. The structure

is simply supported at support beam ends, and the boxes are connected with bar elements modeling the hinges that constrain the vertical displacements at specific points. The team loaded the structure uniformly on the top panels with a pressure of 45 kPa. The goal was to optimize 17 thickness design variables: eight top plates, four bottom plates, three transverse beams and two longitudinal beam webs. Constraints were derived from empirical knowledge captured from previous projects.

The team optimized the structure using two different ini-tial configurations. The first initial configuration set all design variables at their minimum value as defined by the design rules: 7 mm for the bottom plate and 8 mm for the other parts. The other initial configuration set all variables equal to 9 mm. The optimization process ended when the relative change of mass from one iteration to the next was less than 1 percent. Both of the initial design points resulted in similar objective and constraint function convergence. The optimization process took about six hours using a desktop computer with a single core.

The researchers migrated the optimization process to a Techila Technologies Ltd. high-performance computing system cluster. The 18 design points used to create the RSM were run in parallel instead of sequentially, thus significantly reducing the time of each iterative optimization round.

This project demonstrates that design of marine structures can be quickly and inexpensively automated by constructing an RSM based on successive FEA analysis runs. The automated process optimizes the design in much less time than would be required by an analyst performing the same task manually. With this method, MacGregor can respond to a customer inquiry with a speed and accuracy that did not exist before, and with a design that is optimized for each specific application.

Development work has been supported by Finnish Metals and Engineering

Competence Cluster (FIMECC).


W hen a commercial pipeline needs to be repaired, there is no room for second guesses. Industrial oil and

gas pipes can break down due to dam-age, corrosion or both. When they do fail, many thousands of utility customers may be left without service. Furthermore, there could be significant risks, to both the sur-rounding environment and public safety. During such an emergency, composites materials have proven to be critical in making necessary repairs. Repairing per-sonnel require a reliable system that can handle extreme real-world conditions and can be deployed in a timely fashion. For manufacturers of composites pipe repair systems, such as Neptune Research, Inc.

NRI’s repair systems need to perform at harsh ground conditions, including extremes in temperature and precipitation, and underwater.

energy

DEsigning for rEal-WorlD rEpairsLinear and nonlinear structural analyses improve pipeline repair using composites materials.

By Eri Vokshi, Civil Engineer, Neptune Research, Inc., Lake Park, U.S.A.

(NRI), structural analysis from ANSYS plays a critical role in this effort.

At NRI’s Florida headquarters, com-mon practice is to develop virtual proto-types before building physical ones. NRI engineers subject a computational model of a physical project to various load cases to see how the repair system responds. The ANSYS suite of structural

analysis solutions is ideal for such applications, taking into account loads like deformations, vibration charac-teristics and reaction forces. To sat-isfy customers’ expectations, NRI’s repair systems need to perform at harsh ground conditions, including extremes in temperature and precipita-tion, and underwater. In meeting these



ANSYS.COM

� Exposed pipe elbow repaired with FRP composites material

� Pipe hoop stresses in undamaged steel pipe using linear material model (top). The maximum stress is approximately 65,000 psi. Pipe hoop stress modeled using nonlinear material model for steel (bottom), subject to the same load. The maximum hoop stress is approximately 63,000 psi.

performance requirements, NRI designers use ANSYS Composite PrepPost and ANSYS Structural to compare linear and nonlinear analyses of steel pipe and to help determine the effectiveness of repairs using fiber-reinforced polymers.

In many applications of pipe reinforcement, deterioration of pipe-wall material due to corrosion or other physical damage is assumed to be large, which means greater than 50 percent. At such wall deterioration, the designed internal pressure for an undamaged pipe often produces stress that exceeds the yield strength of the remaining steel. Thus, external application of a fiber-reinforced polymer (FRP) composites material is needed to reinforce the weakened section of pipe. Simulation with ANSYS tools has helped the NRI team to predict and verify the perfor-mance of a repair solution. In this case, the repair solution that was analyzed was NRI’s Viper-Skin™ system.

For the nonlinear simulation of a reinforced pipe, the NRI team used ANSYS Composite PrepPost. The steel was modeled as an elastic, perfectly plastic material. The yield and ultimate strengths of the pipe were 43,000 psi and 65,000 psi, respec-tively. Material properties were obtained from the steel-mill cer-tificate (which certifies the manufacturing standards of the mill’s product). For the linear simulation, NRI used an isotropic mate-rial with a defined modulus of elasticity and Poisson’s ratio. For both linear and nonlinear pipe simulations, the team modeled the Viper-Skin composite as an orthotropic material with a linear stress–strain curve. Material properties of Viper-Skin were deter-mined from third-party testing and internal materials testing.

NRI began its simulation study by using ANSYS Structural to analyze an undamaged pipe with an internal pressure of 5,700 psi, which was the burst pressure observed from hydrostatic test-ing. Results from the linear simulation indicated that the inside surface of the pipe was more highly stressed than the outside surface, while the nonlinear simulation showed that the outside surface was more highly stressed. The stresses produced by the burst pressure are beyond the steel’s yield stress, so compar-ing those stresses to the results from a linear stress analysis is not valid. A nonlinear stress analysis is needed for the comparison.

In the next step of the simulation, NRI’s engineers introduced an external defect into the pipe representing 80 percent wall loss while maintaining an internal pressure of 5,700 psi. As part of the repair system, the plan involved filling the physical defect with a proprietary epoxy to optimize load transfer between the

defect and the Viper-Skin. In Composite PrepPost, the team used an isotropic material to represent the epoxy filler. Material prop-erties of the epoxy were determined from third-party testing and internal materials testing. Without repair, ductile yielding would eventually lead to premature rupture of the steel. Simulation pre-dicted that the repair would hold solid even with the 80 percent wall loss, and subsequent physical testing confirmed the validity of the ANSYS model.

The NRI team found that nonlinear analysis capabilities of ANSYS structural mechanics tools combined with Composite PrepPost were very useful in predicting stress distributions in composites-reinforced pipes. The flexibility of ANSYS soft-ware allowed the NRI engineering team to capture the subtle-ties of dealing with the properties of composites materials. Although the research team could perform its job without sim-ulation, ANSYS software gave the typical user the ability both to respond more quickly and to take more details into account. For example, engineers could analyze the predicted behavior


http://www.ansys.com/Products/Simulation+Technology/Structural+Analysis/ANSYS+Composite+PrepPost


Nonlinear analysis capabilities of ANSYS structural mechanics tools combined with Composite PrepPost were useful in predicting stress distributions in composites-reinforced pipes.

� Typical defect profile simulated by NRI engineers (left). Eighty percent of the pipe wall is removed over an area 3.28 inches long by 1.67 inches wide. The simulation results were then compared against physical testing of the same defect machined into a pipe (right).

energy

Simulation gives NRI designers confidence that their repair designs are sound, well before they are put to the test in the real world.

of different composites materials side by side within minutes. Thought of in another way, you could walk to the store and carry groceries home on foot, but driving is faster and allows you to trans-port more items. Similarly, simulation has enabled NRI’s engineers to be more productive.

ANSYS tools have proven to be instru-mental in making sure that demanding oil and gas industry customers obtain the quality repair solutions that they rely on from NRI. In addition to using the soft-ware for detailed linear and nonlinear

analyses, NRI engineers apply har-monic analysis (for determining har-monically time-varying load responses) and spectrum analysis (for random vibrations). Beyond that, the team is working on composites flaw detec-tion models to evaluate the effects of various types of damage and their impact on pipe load capacity.

These types of studies aid a com-pany in reducing its product development cycle while improving performance. This can result in a significant advantage over competitors, and ANSYS simulation tools

have provided NRI with that advantage. Being quick to market is important, but the repair solution also must be easy to use and reliable. More important is that simulation continues to give NRI design-ers confidence that their repair designs are sound — before they are put to the test in the real world.


ANSYS.COM

� Stress distribution in a repaired pipe. Using linear analysis (top), results show the peak stress to be within the flaw, incorrectly suggesting that the repaired pipe would still fail. Using nonlinear analysis (bottom), there is no stress concentration within the flaw area and the pressure load is uniformly distributed, which is a desired condition for pipe repair.

� Actual physical test pipe specimen pressurized until failure. As predicted with ANSYS Composite PrepPost analysis, the repair held solid. This test confirmed the validity of the simulation model and the strength of the composites repair.

40+ World-Class Locations

Network with Leading CAE Professionals

Learn From Your Peers

Visit ANSYS.COM/Conference for more information


http://ansys.com/Conference

tecH tiP

T raditionally, layered composi tes st ruc -tures are modeled as thin structures using shell elements. This approach is valid when

designing thin parts, such as hollow tubes for bikes, panels for airframes and wind turbine blades. But when the parts are more massive, such as gas turbine blades or stringers for pressure vessels, using shell elements is not appropriate. In such cases, both stresses in the direction of the thick-ness and shear stresses out of plane are significant, and solid models are required. Solid models are also appropriate when loads are applied in the direction of the thickness or when the structure is subject to large deformations.

While defining thin-layered composites poses several chal-lenges, the definition of solid composites is even more com-plex. The shapes usually are not simple and require special treatment — a turbine blade, for example. Composites products generally include noncomposites parts that must be included in the simulation. Consequently, the engineering team needs an efficient workflow for the design of products made of lay-ered solid composites and other parts. An effective process starts by examining the layer definition, based on the same method as used for thin structures, then moves on to create solid composites by extrusion. This is followed by the assembly of composites and noncomposites parts, culminating in analy-sis of potential failure of the overall structure.

To highlight this workflow, the example presented is a pres-sure vessel (Figure 1). The entire simulation process is per-formed in ANSYS Workbench (Figure 2) using ANSYS Composite PrepPost (ACP). The workflow begins by defining the geome-try. The model is split into shell composites parts (A) and solid noncomposites parts (D), which are recombined as solids to create the final description of the analysis (B). This combined solid assembly includes connections between parts, loads and boundary conditions, as well as results such as stresses or deformations. The investigation of composites failure occurs as the last step in this process (C).

DEsigning soliD compositEsEmploying ANSYS Workbench workflow streamlines simulation of solid composites.

By Matthias Alberts, CEO, CADFEM US Inc., Greenville, U.S.A., and Pierre Thieffry, Lead Product Manager, ANSYS, Inc.

The entire simulation process is based on a workflow in ANSYS Workbench using ANSYS Composite PrepPost.

� Figure 2. Workflow for design of a composites pressure vessel

� Figure 1. Composites pressure vessel with

titanium caps


http://www.ansys.com/Products/Workflow+Technology/ANSYS+Workbench+Platform



ANSYS.COM

The transfer of the composites material definition in the full assembly is completely automated.

� Figure 3. Definition of ply sequence on inner surface of vessel

� Figure 3b. Surface smoothing using CAD surface (in green)

� Figure 3a. Ply tapering

To start, define the layers on a surface mesh (Figure 3). The surface generally will be the inner or outer surface of the product being designed. Define the lay-ers in the following sequence: defini-tion of materials, fabrics (and, optionally, stackups), orientation of the various sur-faces of the composites (possibly includ-ing draping for highly curved surfaces) and, finally, ply sequence. This approach is very close to the actual manufacturing process. An analogy can be made between the initial surface and a mold, and the ply sequence defined within the simula-tion tool can be the same as the actual fab-ric layup within the mold. However, the simulation tool obviously offers more flexibility in ply ordering, as plies can be swapped easily, modified or removed to achieve the required stiffness, weight and cost requirements.

The critical area in the creation of solid composites is generating a solid model of the layup using a solid extru-sion, based on the previous surface definition of the plies. Advanced capa-bilities, such as ply tapering, surface smoothing or extrusion guidelines, are available to deal with complex shapes (Figures 3a, 3b and 3c). You may apply ply tapering using cutoff rules. Surface smoothing can be performed using the snap-to-geometry feature to fit the extruded model to a given CAD surface. Extrusion guidelines help to extrude the surface model along arbitrary directions.

Another important aspect is han-dling drop-offs. Ply drop-offs can cause damage and delamination in a compos-ite layup. In simulation models, drop-off elements are represented by degenerated brick elements. They usually are made of a homogenous material such as resin.

Once the solid model has been cre-ated, it is automatically merged into the final assembly along with the noncom-posites parts (Figure 4). Automated con-tact detection between parts, loads and boundary conditions, and solution set-tings all can be specified as you nor-mally do for any regular model in ANSYS Mechanical. The transfer of the compos-ites material definition in the full assem-bly is completely automated. Once the model has been solved, standard results such as deformations or stresses can be displayed on the full model.

Additional capabilities are available to analyze potential failure of the compos-ites parts. Three-dimensional failure cri-teria (maximum stress, maximum strain, Tsai–Wu, Tsai–Hill, Puck, Hashin, Cuntze) are available. A typical failure plot gives the analyst information on the risk of

failure and the potentially problem-atic elements and layers in the assembly (Figure 6). You can even create a graph to show failure criteria values through all of the layers at a given location on the model.

Employing the Workbench-based workflow for solid composites delivers benefits from additional capabilities. If the composites parts are subject to pres-sure from a flow environment, you can easily add the fluid flow simulation to the simulation, and pressures can be mapped automatically on the struc-tural model. Users also have the abil-ity to parameterize a model to perform sensitivity or optimization studies based on geometry or composites variations (thicknesses, orientation, etc.) using the failure criteria as a performance indica-tor of the design.


� Figure 3c. Extrusion guidelines � Figure 4. Assembled model showing composites and noncomposites parts (metallic caps in light gray)

� Figure 6. Post-processing composites showing failure on full model (left) or through element layers (right) in charts

� Figure 5. Stresses on assembled model

tecH tiP


ANSYS.COM

centrifugal comPreSSorS

Dresser-Rand designs compressor stages to operate at higher fl ow coeffi cients and higher machine or inlet-relative Mach numbers.

Pushing the Envelope

cfD simulation contributes to increasing the operating envelope of a centrifugal compressor stage.By James M. Sorokes, Principal Engineer; Jorge E. Pacheco, Manager, Aero/Thermo Design Engineering; and Kalyan C. Malnedi, Manager, Solid Mechanics Group, Dresser-Rand Company, Olean, U.S.A.

c entrifugal compressors, also called radial compressors, play a critical role in many process

industries, including oil and gas, petro-chemical, and gas transmission. These machines are used to compress a gas or a gas–liquid mixture into a smaller volume while increasing its pressure and temperature.

comprEssor DEsign challEngEsProcess industries are looking for smaller-footprint compressors for space-sensi-tive applications, such as off shore, subsea and compact plant designs. Dresser-Rand reduces compressor footprints by design-ing stages to operate at higher flow coeffi cients and higher machine or inlet-relative Mach numbers. The company is among the largest global suppliers of rotating equipment solutions for long-life, critical applications.

In recent years, the industry has placed greater emphasis on achieving a wide operating range so that, for exam-ple, compressors can handle a wider range of fl ow rates at diff erent stages of a well’s lifecycle. Engineering simulation is an important tool in addressing these market challenges. Dresser-Rand has been using ANSYS CFX software since the 1990s to develop many new compressor designs for process industries and other applications.

Two factors limit the overall operating range of a compressor: surge or stall mar-gin, and overload capacity. Surge or stall margin limits the compressor’s ability to operate at fl ow rates lower than design, while overload capacity limits the ability to operate at higher rates. Rotating stall arises when small regions of low momen-tum or low pressure (referred to as stall cells) form in the fl ow passages and begin to rotate around the circumference of the compressor. These fl ow and/or pressure


http://www.ansys.com/Products/Simulation+Technology/Fluid+Dynamics

� Centrifugal compressors operate by adding velocity pressure or kinetic energy to the fluid stream and then converting that kinetic energy into potential energy in the form of static pressure. Kinetic energy is added by rotating impellers, while the conversion of velocity pressure to static pressure occurs in downstream stationary components such as diffusers, return channels and volutes.

disturbances cause unbalanced forces on the compressor rotor, leading to unwanted vibration issues and reduced compres-sor performance. Surge occurs when the compressor is no longer able to overcome the pressure in the downstream pip-ing and pressure vessels, and the flow is forced backward through the compressor.

For most centrifugal stages that oper-ate at high inlet-relative Mach numbers, low-momentum regions can form along the shroud side of parallel-wall vaneless diffusers. Typically the size of this region increases as flow is reduced until diffuser stall results. In developing a new high-head stage for a high–Mach number com-pressor, the Dresser-Rand team observed an interesting phenomenon both in com-putational fluid dynamics (CFD) simula-tion and test results: A sudden migration

of the low-momentum region from the shroud side to the hub side of the diffuser occurred as the flow rate reduced, just prior to stall [1]. The impeller used in the study is operated over a machine Mach number range of 0.85 to 1.20. The initial design had a vaneless diffuser that was pinched at the shroud and then followed by a parallel wall section. In analyzing test results, engineers established that the shift of the high-momentum region occurred much earlier for this high-head stage than for lower-head stages. As a result, the surge margin was significantly lower than low-head stages, an unac-ceptable drop in operating range. Since the stationary components were stalling before the impeller due to low momentum shift, the team decided to use CFD to opti-mize the diffuser and return channel.

using cfD to optimizE thE DEsignDresser-Rand engineers conducted all analyses using ANSYS CFX software for a sector model that included the upstream inlet guide vane, impeller, diffuser, return bend, return channel and exit section. In this case, the grid was composed of more than 5 million total elements using a tet-rahedral mesh with wedge elements for the boundary layers. Engineers modeled the interfaces between stationary and rotating components using a stage inter-face that performs a circumferential aver-aging of the fluxes through bands on the interface. The k-epsilon turbulence model and a high-resolution discretization scheme were used.

The team evaluated several combi-nations of pinch, shroud-tapered and



ANSYS.COM

� Velocity profile at 90 percent flow for the optimized stationary component design shows a much smaller low-momentum region.

� Absolute velocity flow angle at the diffuser exit for the original design shows high tangential-flow angles, indicative of low-momentum flow that often leads to formation of stall cells.

� Absolute velocity flow angle at the diffuser exit for the optimized design shows greatly reduced high tangential-flow angles.

Inlet

Inlet guide vane

Impeller

Hub

� Velocity profile at 90 percent flow for the original stationary component design shows a large low-momentum region at the hub side in the vaneless diffuser and return bend.

Return bend

Vaneless diffuser

ShroudReturn channel vanes

hub-tapered diffusers. Engineers iter-ated to a diffuser design that is pinched and tapered on both hub and shroud sides to significantly reduce low-momentum regions that were forming on either side of the diffuser exit at low flow. They reduced the return channel width and redesigned the return channel vanes to match the new flow incidence. CFD results showed that the new design significantly reduced the size of the low-momentum region in the diffuser and return channel. It also

considerably delayed the shift of the low-momentum region from the shroud side to the hub side, delaying the onset of stall.

Comparison between the original and optimized designs shows a substan-tial reduction in the absolute velocity flow angle relative to the radial line at the diffuser exit plane. Highly tangential flow angles greater than 75 degrees gen-erally are indicative of very low momen-tum, which leads to formation of stall cells in stationary components. The pressure

recovery plots for both original and opti-mized geometries show that the optimized geometry has lower pressure recovery on the overload side but better performance on the surge side. The lower recovery at overload for the optimized geometry is most likely due to the narrow station-ary component passages, which results in higher gas velocities and lower pres-sure recovery. However, this geometry also contributes to improving the flow in the stationary components, resulting in


Structural AnalysisDresser-Rand structural engineers optimize the impeller design to keep the static stresses both below those seen in similar families of impellers and below allowable mate-rial yield strengths. The lower the stresses, the faster the impeller can be run. During the design process, engineers also analyze the design to see if there are any possible res-onance issues caused by upstream or downstream station-ary components.

Most of the time, structural damages to the impellers are due to mechanical fatigue. Dresser-Rand follows the in-house dynamic audit process [3] to evaluate the fatigue life of the impellers. The dynamic audit process involves a series of successive analysis runs, starting with a modal analysis and plotting of a SAFE diagram [4] for identifying interfer-ences. This is followed by harmonic response analyses to compute dynamic stress levels in an impeller at identifi ed SAFE interferences. A minimum factor of safety is then com-puted for all locations in the impeller based on the static and dynamic stresses, material properties and the construction method used for that impeller. The structural team has auto-mated much of the structural design process by writing APDL macros and FORTRAN programs, which have reduced sim-ulation time from more than a week to one to two days per design iteration.

� Typical steady-state plot of impeller

� Industrial centrifugal compressor courtesy Dresser-rAnD.

Tests validated CFD simulation prediction of about 10 percent improvement in the surge margin of the new design.


better pressure recovery at lower fl ow. The CFD results predicted an improvement in surge margin of approximately 15 percent.

Tests validated CFD simulation prediction of an improve-ment of about 10 percent in the surge margin of the new design. Flow angle measurements at the diff user inlet, diff user exit and return channel inlet confi rmed the CFD prediction of a delay in the low-momentum shift. Further, the redesign was successful in maintaining the same head and effi ciency levels as the pre-vious design had. About 1.5 percent of overload margin was lost due to the reduced passage areas in the stationary compo-nents. However, this was deemed acceptable, as the stage is not expected to be operated at high fl ow levels close to choke.

Impellers are subjected to inlet and exit flow variations through the stage, and therefore they must be designed to with-stand the alternating pressure loads due to these variations in addition to withstanding steady loads. The structural team used ANSYS Mechanical software and in-house tools to ensure that the new design meets the static and dynamic stress requirements.

Dresser-Rand’s use of CFD simulation to optimize the sta-tionary components of a new centrifugal compressor design accomplished several goals: This new design delayed the trans-



ANSYS.COM

� The optimized design shows better pressure recovery on the surge (left) side of the pressure recovery curve.

� CFD results and physical tests provide similar estimates of compressor efficiency.

� Testing shows that optimized design improves the operating range.

Performance Parameter Original Design Optimized Design

Normalized polytropic efficiency at design flow 1.000 1.001

Normalized polytropic head coefficient at design flow 1.000 1.003

Surge margin 6.1 16.0

Overload margin 13.4 12.1

Dresser-Rand delivered a highly efficient compressor with a wide operating range in a small footprint.

fer of the low-momentum zone from the shroud side of the diffuser to the hub side, and it shows how proper sizing of station-ary components in the early stages of the design process can increase the compres-sor’s operating range. The end result is that Dresser-Rand delivered a highly effi-cient compressor with a wide operating range in a small footprint [2].

References[1] Sorokes, J. M.; Pacheco, J. E.; Vezier, C.; Fakhri, S.

“An Analytical and Experimental Assessment of

a Diffuser Flow Phenomenon as a Precursor to

Stall”. Proceedings of ASME Turbo Expo 2012, 2012,

Volume 8: Turbomachinery, Parts A, B and C.

[2] Fakhri, S.; Sorokes, J. M.; Vezier, C.; Pacheco,

J. E. “Stationary Component Optimization and

the Resultant Improvement in the Performance

Characteristics of a Radial Compressor Stage”.

Proceedings of ASME Turbo Expo 2013, 2013.

[3] Schiffer, D. M.; Syed, A. An “Impeller Dynamic

Risk Assessment Toolkit”. Proceedings of the 35th

Turbomachinery Symposium, 2006, pp. 49–54.

[4] Singh, M. P.; Vargo, J. J.; Schiffer, D. M.;

Dello, J. D. “SAFE Diagram — A Design and

Reliability Tool for Turbine Blading”. Proceedings of

the Seventeenth Turbomachinery Symposium, 1988,

pp. 93–101.


Raising the Standardsfluid-mechanical simulation can help prevent offshore disasters by supporting development of more effective structural standards.By Richard Grant, President, Grantec Engineering Consultants Inc., Halifax, Canada

tHougHt leader

the critical nature of off shore structures has been tragically demonstrated by incidents such as the Piper Alpha explosion in the North Sea in 1988, which took 167 lives, and the Deepwater Horizon fi re in 2010 in the Gulf of Mexico, which caused

both loss of life and environmental damage. Every off shore disaster provides an opportunity to understand its causes and to ensure that future structures are designed to avoid its recur-rence. Computer simulation can play an important role by helping to diagnose real-life or potential disasters and evalu-ate the eff ectiveness of alternate remediation methods.

Off shore oil and gas production in Canada started in the late 1980s with the development of the Cohasset and Panuke fi elds (fi rst oil in 1992) off the coast of Nova Scotia, followed by the Hibernia fi eld (fi rst oil in 1997) off Newfoundland. Atlantic Canada’s off shore presents one of the harshest weather envi-ronments in the world. Oil-related tragedies already have occurred in this area, such as the 1982 sinking of the Ocean Ranger mobile off shore drilling unit, with complete loss of life at the Hibernia fi eld.

avoiDing rEpEat mistaKEsMany of the off shore disasters that occurred throughout the world could have been prevented if only the best practices

Richard Grant has worked on developing structural standards for off shore platforms since 1997, when he became a found-ing member of the Canadian Advisory Committee (CAC) on Off shore Structures Standards under the Standards Council of Canada (SCC). His involvement with the CAC began with pro-

viding input into the off shore structural standards then being developed by the International Organization for Standardization (ISO). Shortly afterward, while work-ing on a Canadian off shore project, Grant noted serious shortcomings in Canadian regulations and standards pertaining to fi re and explosion safety. He has since been instrumental in correcting defi ciencies in Canadian off shore standards through his work with the Canadian Standards Association (CSA). He was subsequently called upon by the international community to assist with the related work being performed under ISO.


ANSYS.COM

available at the time had been followed. During develop-ment of the Sable and Terra Nova offshore projects in Atlantic Canada, it was recognized that Canada’s CSA offshore stan-dards needed to be reviewed and, where necessary, updated. It was also recognized that Canada’s efforts in this area would be better served through direct participation in developing the new offshore structures standards being established under ISO. International standardization of industry best practices helps ensure that lessons learned are captured, and past mis-takes are not repeated.

The offshore environment’s nature means that realistic physical testing is often too expensive or dangerous and, in many cases, is simply impossible. Simulation plays a crucial role by enabling those involved in developing standards, as well as those designing offshore structures, to evaluate poten-tial disaster scenarios and determine the impact of imple-menting requirements to help improve the safety of workers and the environment.

ship-to-platform collisionANSYS Mechanical structural simulation has been used to gain a better understanding of one of the more dangerous offshore scenarios: a ship-to-platform collision. The 2005 incident at Bombay (Mumbai) High North, off the coast of India, is a good example for illustrating this situation. A vessel was called to the platform to transport an injured person to shore for med-ical attention. The vessel came too close to the platform, and it hit and ruptured risers carrying gas to the platform. The subsequent fire destroyed the platform and resulted in many fatalities.

To avoid such catastrophes, offshore structures are required to safely protect critical components such as risers and to absorb energy during a collision. This collision energy is dependent on the vessels permitted within the safety zone around the platform. In performing a simulation of this type, a structure is typically modeled using shell elements with nonlinear material properties and large displacements to accurately represent the resistance of the structure. Collision causes tubular denting, leading to large plastic strains. The structure absorbs energy as its tubular members are crushed by the impact of the vessel. Analysis can accurately capture the amount of energy that the structure can absorb. This type of simulation is invaluable to assess the safety of offshore platforms and can be used to assure that they can withstand

certain types of collisions without catastrophic failure, as required by the standards.

procEss prEssurE vEssEl intEgritYGrantec is performing research pertaining to the integrity of process vessels subjected to external hydrocarbon gas explo-sions. This research is aimed at better understanding the

Progress in Canada’s Offshore StandardsCanada’s own offshore standards were significantly improved in the area of fire and explosion safety through new provisions in the CSA Offshore Structure Standards (2004). With Canadian input, similar pro-visions were developed for the ISO 19901-3 Topsides offshore structures standard. The ISO standards will improve safety in offshore structures around the world. Canada also holds leadership positions in ice loading and concrete construction requirements for offshore structures; the country’s experts have provided signifi-cant input in these areas of the ISO standards.

Lessons learned from the international community and from Canadian projects have been considered in the rewrite of the Canadian offshore structures weld-ing requirements, contained in the CSA W59-13 weld-ing standard, which is referenced by the ISO 19902 standard for Fixed Steel Structures. Canada’s par-ticipation with international standards bodies bene-fits Canada by ensuring that the country’s standards reflect the advances made by the international commu-nity; it also helps to ensure that advancements made in Canada are reflected in international standards. Many of the new ISO offshore structures standards have been adopted as National Standards of Canada, replacing the CSA S47x Canadian offshore structures standards.

Analysis is used to confirm that the structure can absorb sufficient energy to withstand impact from ships permitted within the safety zone.

Simulation can play an important role by helping to diagnose real-life or potential disasters and evaluate the effectiveness of alternate remediation methods.



gas leak that, in turn, produced an explosion that tore loose a bulkhead. The explosion launched the bulkhead, which then ruptured process piping and equipment, resulting in a fire that continued to escalate and destroy the platform with sig-nificant loss of life.

Process vessel integrity research at Grantec is being per-formed using multiphysics simulation software from ANSYS. The simulation is a fully coupled transient fluid–structure interaction (FSI) analysis that uses ANSYS Mechanical for the structural simulation and ANSYS CFX for the computational fluid dynamics (CFD) analysis. The vessel shell and saddle supports are modeled with shell elements, and the support structure is modeled using beam elements to simulate deck flexibility. Nonlinear material properties and large deflec-tions are also incorporated. A fluid domain outside the vessel is used for simulating the transient explosion acting on the external surface of the vessel, and an internal fluid domain is used for the fluids (liquid and gas) in the vessel. The time his-tory of the explosion is applied at the inlet of the fluid domain upstream of the vessel. The explosion, advancing rapidly through the domain, results in high transient drag loading on the vessel, causing it to move. The movement of the ves-sel causes the internal liquid to undergo sloshing, generating additional loads on the shell of the vessel.

Grantec engineers leverage the ANSYS HPC (high-perfor-mance computing) multi-processor option to significantly speed up the computationally demanding analysis. The com-pany migrated from two HPC licenses to the HPC Pack license, which resulted in a four-times speed improvement for multi-physics simulations on an HP Z820 workstation. The team is looking at ways to further improve turn-around on multiphys-ics simulations — such as using GPU capabilities, additional hardware and other methods. The results of the research conducted will be used to assess and guide future standards requirements. International cooperation between standards-making bod-ies helps to substantially improve offshore platform standards in the areas of structural integrity, control and mitigation of accidents, and protection of safety-critical systems. Simulation enables engineers to understand and diagnose many poten-tial accidents and evaluate the effects of possible design stan-dards in protecting human life and the environment.

The nature of the offshore environment means that realistic physical testing is often too expensive or dangerous, and in many cases is simply impossible.

tHougHt leaderSHiP

dynamic response of the process vessel shell (pressure enve-lope), including effects due to internal fluids and the support structure, such as saddles, process skids and decks. Rupture of the pressure envelope of a hydrocarbon process vessel, or the failure of vessel supports, during a hydrocarbon explosion can cause the event to escalate. In the case of a rupture, more hydrocarbons are released to fuel a fire, and, in the case of support failure, the vessel could become a projectile impacting and impairing other safety-critical systems. In the Piper Alpha disaster, failure of hydrocarbon process piping and equipment resulted in the escalation of the fire onboard the platform. In this incident, a loose blind flange on piping resulted in a

Simulation of accidental ship collision with wellhead platform

Sloshing inside vessel caused by a hydrocarbon explosion




http://www.ansys.com/Products/Workflow+Technology/High-Performance+Computing

ANSYS.COM

Fluid–structure interaction simulates an explosion that induces load on pressure vessel. Image shows instantaneous velocity vectors of air moving as a result of explosion.

Time slice shows streamlines of explosion flow over pressure vessel.


Within the global energy supply chain, design, engineering and manufacturing

groups span multiple geographies and involve teams engaged in discovery, generation, collection, storage, transportation, distribution and more. Each sector works on a broad set of challenges, solving problems that involve different physics, scale and components. Beyond simulation software, the ANSYS network of technical experts works with oil and gas customers around the world. We operate from local offices close to energy companies in Houston, Aberdeen, Rio de Janeiro, Stavanger, Kuala Lumpur, Calgary, Moscow and more. With our network of channel partners, the company fosters close relationships with customers and provides local value added service and support.

ANSYS: A Platform for Global Collaboration

Contacts and Locationsansys.com/About+ANSYS/ContactsLocations

http://www.ansys.com/About+ANSYS/ContactsLocations

ANSYS, [email protected]

ANSYS is dedicated exclusively to developing engineering simulation software that fosters rapid and innovative product design. Our technology enables you to predict with confi dence that your product will thrive in the real world. For more than 40 years, customers in the most demanding markets have trusted our solutions to help ensure the integrity of their products and drive business success through innovation.

© 2014 ANSYS, Inc. All Rights Reserved.

ANSYS, Inc. does not guarantee or warrant accuracy or completeness of the material contained in this publication.

ANSYS, BladeModeler, CFD, CFX, DesignerSI, DesignModeler, DesignXplorer, Fluent, HFSS, ICEM CFD, Mechanical, Multiphysics, Realize Your Product Promise, Sentinel, Simulation-Driven Product Development, Structural, Workbench, and any and all ANSYS, Inc. brand, product, service, and feature names, logos and slogans are registered trademarks or trademarks of ANSYS, Inc., or its subsidiaries located in the United States or other countries. All other brand, product, service, and feature names or trademarks are the property of their respective owners.

Oil and gas operations are complex and dangerous. One mistake and failure can destroy life, capital and reputations overnight.

To mitigate that risk, oil and gas industry leaders use ANSYS engineeringsimulation technology to meet equipment and process requirements virtually,since real-life testing in new oil and gas fi elds can be cost-prohibitive.

To learn more please visit our website www.ansys.com/energy

Realize Your Product Promise®

Drilling to New Depths with ANSYS

http://www.ansys.com/Industries/Energy

http://www.ansys.com/

mailto:[email protected]

ansys advantage special issue: oil and gas/media/ansys/corporate/... · · 2017-04-05in a drill...

Documents